Semantics (was Re: Inheritance in XML [^*])

[len bullard]

Tim Bray wrote:
> 
> At 01:15 PM 4/23/98 -0400, Paul Prescod wrote:
> >If XML had no semantics, then XSL, XLL and the DOM would have to
> >explicitly describe the mapping from syntactic features to the abstract
> >nodes that they work on. But they do not, because XML has semantic
> >concepts like "element, "element type", "notation" and "attribute" that
> >are *described by* the syntax.
> 
> Well, we just have a difference of perception. 

Yes.  OTOH...

> Dan Connolly has argued repeatedly and forcefully that
> the spec could be completely re-written to avoid discussion of the
> processor's actions (he is right) and that this would be an improvement
> (I'm not convinced).

That a spec can be rewritten to to exclude information of 
use is not arguable.  Yet contextually, the project was/is 
SGML On The WEB.  There are two main requirements in that 
simple title:  

a. SGML (done)  

b. On The Web.  b is a systemic requirement.  

Systemic requirements require semantic information.  Removing 
those from the spec unlimits the spec, but I think makes them 
much less useful.  Failure to do this for SGML systems   
made SGML less useful, and arguably, less attractive 
to implement.  Why?  Friendly to Information Maintenance but not 
very good for system interoperability.  The industry did not 
get enough cohesion among systems to get community reinforced 
growth. No amplification.  

It depends on what one needs to improve.  One 
can get hung up on information purity issues and not see 
the systemic problems of a standard framework.  Isn't avoiding 
that mistake precisely what made HTML/HTTP work?

> The fact that the XML processor has a couple of required *behaviors*,
> most notably error handling, does not constitute anything like
> what I think of in connection with the term "semantic".

They are semantics/behaviors defined to make them useful.

> But in the instance, Elements and attributes don't mean anything in and
> of themselves.  They doubtless have semantics that are used by humans and
> computer programs in particular application domains, but that's none of
> our business.

Yes they do.  They are meant to readable to humans and machines.  This 
is a contractual semantic.  The meaning is in the intent of the design.  
Say 'does this' and the semantic is there in the requirement.

So?

> And finally... words are only of use in facilitating human
> communication when there is some shared understanding as to their
> denotation and connotation.  The term "semantic", judged by this standard,
> has clearly and empirically lost its usefulness in this discussion.

Hmmm... no.  But the semantics of markup systems (not markup) are 
limited to the agreed upon meanings.  You have contractual semantics 
(design requirements) and systemic semantics (environment where 
requirements are testable).

> But of this I am confident: elements, attributes, and entities don't
> mean anything in and of themselves. -Tim

They mean what we agree to make them mean.  Right now, we agree 
to make them readable to human systems and machine systems.

XML is a system-centric standard.  SGML is not.  XSL, 
DOM, XLL are system applications.  This does not mean they 
will not meet the loftier goals of information evolution, 
but they have to operate to do that.  

The pursuit of meaning is a human pursuit.  Machines, like XML, 
don't care.

len

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)